Compilation of narrowband spectral detection systems for linear MIMD machines

The author discusses the design of a program that maps a class of digital signal processing systems, called narrowband spectral detection systems, to linear MIMD machines. Such systems contain a mixture of data-parallel, systolic and purely serial computations. He describes a new technique, called geometric scheduling, that exploits the special features of the first two styles of computation, and that can also incorporate tasks that are neither data-parallel nor systolic. The resulting schedules contain all necessary communication code, which is automatically generated. This paper includes performance figures for this method on a typical narrowband spectral detection system.<<ETX>>

[1]  H. T. Kung,et al.  The Warp Computer: Architecture, Implementation, and Performance , 1987, IEEE Transactions on Computers.

[2]  Edward A. Lee,et al.  Static Scheduling of Synchronous Data Flow Programs for Digital Signal Processing , 1989, IEEE Transactions on Computers.

[3]  Shekhar Y. Borkar,et al.  iWarp: an integrated solution to high-speed parallel computing , 1988, Proceedings. SUPERCOMPUTING '88.

[4]  Jeffrey D. Ullman,et al.  NP-Complete Scheduling Problems , 1975, J. Comput. Syst. Sci..

[5]  W. Daniel Hillis,et al.  Data parallel algorithms , 1986, CACM.

[6]  H. T. Kung Why systolic architectures? , 1982, Computer.